Abstract: The energy supply devices of traditional tracked vehicle braking systems are mostly composed of hydraulic components. This braking method has problems such as complex hydraulic pipeline wiring, delayed braking, and easy hydraulic oil leakage. With technological breakthroughs in fields such as artificial intelligence and new energy, the braking systems of tracked vehicles are also developing towards electrification and intelligence. Among them, pure electric mechanical braking systems, represented by electronic mechanical braking systems, are an important development direction for future braking systems due to their advantages of simple structure, fast response, and high control accuracy. This article takes the braking system of a certain type of tracked vehicle as the research object, conducts research on the inertia load characteristics of mechanical braking, designs a complementary filter method suitable for acceleration estimation, optimizes and complements the estimation results of model prediction method and differential method, synthesizes the advantages of the two methods, improves the dynamic response quality of the estimation results, and compensates for the robustness of the system. Through model simulation and result verification, this method has better dynamic performance and stability during the braking process, providing accurate and reliable acceleration signals for inertial load compensation.